IFNc sensitization to TRAIL-induced apoptosis in human thyroid ...

Oncogene (2004) 23, 928–935

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $25.00 www.nature.com/onc

IFNc sensitization to TRAIL-induced apoptosis in human thyroid carcinoma cells by upregulating Bak expression Su He Wang1,3, Emese Mezosi1, Julie M Wolf1,3, Zhengyi Cao1,3, Saho Utsugi1, Paul G Gauger2, Gerard M Doherty2 and James R Baker Jr*,1,3 1 Department of Medicine, University of Michigan Medical Center, Ann Arbor, MI, USA; 2Department of Surgery, University of Michigan Medical Center, Ann Arbor, MI, USA; 3Center for Biologic Nanotechnology, University of Michigan Medical Center, Ann Arbor, MI, USA

TRAIL preferentially induces apoptosis in tumor cells and virus-infected cells. Unlike other tumor necrosis factor family members, TRAIL does not kill cells from most normal tissues and has thus been proposed as a promising new cancer treatment. Our study demonstrated that IFNc combined with TRAIL can trigger apoptosis in vitro in several resistant thyroid tumor cell lines, such as thyroid anaplastic carcinoma cells (ARO cells), while either agent alone exerts only a minimal effect. We further tested this effect on a mouse thyroid tumor model, when in vivo tumor growth was also significantly inhibited by this combination. The mechanism of how IFNc sensitized thyroid carcinoma cells to TRAIL-induced apoptosis was investigated by screening global gene alterations in ARO cells treated with IFNc. Microarray data revealed that a proapoptotic gene, Bak, is markedly upregulated by IFNc, and this was confirmed by RNase protection assay. Western blot analysis also showed a significant increase in Bak at the protein level. Upregulation of Bak and sensitization for apoptosis by IFNc was blocked by overexpression of antisense Bak in ARO cells. Furthermore, overexpression of Bak sensitized ARO cell to TRAIL-induced apoptosis without the need for IFNc pretreatment. This suggests that Bak is a regulatory molecule involved in IFNc-facilitated TRAIL-mediated apoptosis in thyroid cancer cells. Oncogene (2004) 23, 928–935. doi:10.1038/sj.onc.1207213 Published online 1 December 2003 Keywords: Bak; TRAIL

thyroid

cancer;

apoptosis;

IFNg;

Introduction Anaplastic thyroid carcinoma (ATC) evolves when neoplastic thyroid cells lose differentiated features. Resistance of ATC to most therapies still remains a major concern in the treatment (Ain, 1998), and efforts to develop new approaches for ATC therapy have not produced promising results. For example, a study of *Correspondence: JR Baker Jr; E-mail: [email protected] Received 7 August 2003; revised 2 September 2003; accepted 12 September 2003

paclitaxel against ATC showed that this drug could not limit mortality of their malignancy (Ain et al., 2000). The effect of manumycin, a farnesyl : protein transferase inhibitor, alone and in combination with paclitaxel, also has been studied in anaplastic thyroid carcinoma cells (ARO cells). Unfortunately, there was no synergic effect in the combination of manumycin and paclitaxel against ARO cells in vivo (Yeung et al., 2000). A study of the combination of cisplatin and gemcitabine against anaplastic thyroid cancer cell lines has concluded that the interaction between these two drugs might relate to DNA-synthesis inhibition and S phase arrest (Voigt et al., 2000), but these studies have not been confirmed in vivo. Thus, the current therapeutic options for dedifferentiated thyroid carcinoma are unsatisfactory. Although differentiated thyroid cancer cells still have the ability to concentrate iodine, in some thyroid cancers this capacity is decreased (Braga-Basaria and Ringel, 2003). Consequently, these differentiated thyroid cancers respond poorly to radioiodine treatment. Novel therapies for ACT and radioiodine-irresponsive differentiated thyroid cancers are vital. Recently, increased attention has been focused on the role of apoptosis in mediating the cytotoxic effects of anticancer agents (Fisher, 1994). The induction of apoptosis has been shown to be an important determinant of the response of most tumors to cytotoxic therapy. Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), a newly identified member of the TNF superfamily, is well recognized to initiate apoptosis by interacting with cell surface death receptors 4 and 5 (DR4 and DR5) (Kayagaki et al., 1999). The signal is propagated through caspases 8 and 10, finally leading to activation of effector caspases such as caspase 3. This pathway is regulated at several levels, including expression of decoy receptors, inhibitors of death-inducing signaling complex, caspase inhibitors and bcl-2 family members (Strasser et al., 2000). Recent studies showed that TRAIL may have particular importance as a therapeutic agent for some cancers, as it produces selective toxicity toward neoplastic cells as opposed to normal cells (Pan et al., 1997; Bonavida et al., 1999). Therefore, TRAIL is a potential novel therapy for ATC. Although it has been reported that thyroid carcinoma cells are sensitive to TRAIL (Mitsiades et al., 2000), we

IFNc sensitizes thyroid carcinoma cells to apoptosis SH Wang et al

929

have found that TRAIL induces apoptosis in most thyroid cancer cells only when regulators that block cell death are inhibited (Bretz et al., 1999a, b). Previous studies from our laboratory have shown that inflammatory cytokines, such as IFNg and TNFa, are involved in the regulation of the death pathway of thyroid follicular cells both in vitro and in vivo (Bretz et al., 1999a, b; Wang et al., 2002). Recently, IFNg has been shown to promote apoptosis in thyrocytes of an IFNg-transgenic animal (Caturegli et al., 2000). Additionally, IFNg modulates TRAIL-mediated apoptosis in human breast and colon tumor cells (Langaas et al., 2001; Ruiz-Ruiz and Lopez-Rivas, 2002). An interesting study has demonstrated IFNg sensitization of TRAIL-induced apoptosis by upregulating caspase 8 in neuroblastoma (Fulda and Debatin, 2002); however, caspase 8 was not regulated by IFNg in thyroid cancer cells. This may indicate the regulation of caspase 8 by IFNg in neuroblastoma is a cell type specific. The mechanism of the IFNg effect on TRAIL-induced apoptosis in human thyroid cancer cells has not been clearly studied. In the present study, we investigated the mechanism of TRAIL-induced apoptosis facilitated by IFNg. Our data demonstrate that IFNg significantly increases the sensitivity of human thyroid cancer cells to TRAILmediated apoptosis in vitro as well as in vivo, and its effect involved the upregulation of proapoptotic Bak. In contrast, normal human thyroid cells were not responsive to the IFNg sensitizing effect and this was correlated to concentration of Bak, which was not altered by IFNg in the same time frame. The differential action of IFNg and TRAIL on apoptosis in neoplastic and normal thyroid cells may have clinical importance.

Results Sensitization of ARO cells by IFNg to TRAIL-induced apoptosis Primary thyroid cancer cells and the ARO cell line did not demonstrate apoptosis in response to recombinant human TRAIL. However, after IFNg pretreatment, ARO cells underwent apoptosis in response to TRAIL (Figure 1a). Similar results were also observed in the human follicular thyroid carcinoma cell line (FRO cells) and human papillary thyroid carcinoma cell line (NPA cells) (Figure 1c). The response to apoptotic stimuli after IFNg treatment was specific for thyroid cancer cells, as primary normal thyroid cells remained unresponsive to TRAIL even after pretreatment with IFNg (Figure 1b). ARO cell sensitization to TRAIL-mediated apoptosis by IFNg was completely blocked by addition of sDR5 (Figure 1d). These experiments were performed three times with consistent results. Comparison of the effect of IFNS on TRAIL-induced apoptosis Dose–response studies of TRAIL-induced apoptosis indicated that 100 U/ml of IFNg were required for

Figure 1 Sensitization of ARO cells by IFNg to TRAIL. ARO cells (a) and normal thyroid cells (b) were pretreated with 100 U/ml of IFNg for 24 h, then treated with a serial dose of recombinant TRAIL. FRO and NPA cells were treated in the same condition (c). After IFN pretreatment, some ARO cells were treated with sDR5 and TRAIL (d). Cell death was determined by staining cells with FDA and PI, and analysed by flow cytometry. Data are presented as the mean7s.d. percent cell death of triplicate measurements

maximal effect on the response of thyroid cancer cells to apoptotic stimulus (Figure 2a). Although signaling crosstalk between IFNs has been reported, type I IFNs Oncogene


930

(IFNa and IFNb) had less ability to sensitize ARO cells to TRAIL-induced apoptosis compared to IFNg (Figure 2b).

Table 1

Alteration of Bcl-2 members detected by microarray Bcl-2 family (IFNg/untreated)

Gene Fold changes

BAK 4.6

BAX 0.4

BCL2 1

BCLx 0.89

BID 0.55

BIK 1.9

Effects of IFNg on gene expression in ARO cells We first used a cDNA microarray to search for candidate molecules involved in the cellular response to IFNg. The array data showed that a proapoptotic protein, Bak, was upregulated 4.6-fold by IFNg, and other members of the Bcl-2 family were also regulated (Table 1). To confirm the microarray findings, RNase protection assay was used to quantify RNA expression of the Bcl-2 family members. Bak mRNA was significantly increased by IFNg (Figure 3a), which was consistent with the result of the microarray assay, whereas Bcl2, Bclx, Bax and Bid mRNA levels were not affected. This also revealed dose-dependent expression of Bak, with a maximal expression found at a concentration of 100 U/ml IFNg, which is also the maximal effect concentration of IFNg for sensitizing thyroid cancer cells to TRAIL-induced apoptosis. Western blot analysis showed that Bak protein level was significantly increased by 100 U/ml of IFNg

Figure 2 Comparison of the effects of two types of IFN in sensitizing thyroid cancer cells to TRAIL-induced apoptosis. (a) Dose–response of IFNg sensitization of ARO cells to TRAIL was performed: cells were pretreated with IFNg (0.1–1000 U) for 24 h, and then incubated with 1000 ng/ml of TRAIL. (b) ARO cells were pretreated with IFN a, b and g, and then incubated with TRAIL. Cell death was determined by flow cytometry Oncogene

Figure 3 Analysis of mRNA and protein concentrations for Bcl-2 family members. (a) RNase protection assay was performed on RNA isolated from the untreated ARO cells (lane 1) and IFNg-treated ARO cells: 10 U (lane 2), 100 U (lane 3) and 1000 U (lane 4). Messenger RNA for Bcl-2 family members was detected. GAPDH was used as a control to standardize RNA concentrations. (b). Effects of IFNg on Bak protein concentrations in thyroid cancer cells. ARO cells were incubated with 100 U/ml of IFNg for 24 h, and protein (10 mg in each lane: lane 1, untreated; lane 2, IFNg; lane 3, TRAIL; lane 4, IFNg þ TRAIL) was used for analysis of Bak. A 24-kDa band of Bak protein and a 42-kDa band of actin were detected by Western analysis using monoclonal antibodies to human Bak and actin

treatment and that TRAIL did not affect Bak expression (Figure 3b). We also demonstrated that Bak protein expression in normal thyroids cells did not show an


931

increase by 24 h of IFNg treatment (data not shown). These experiments were performed three times with consistent results. Activation of mitochondrial pathway in ARO cell apoptosis induced by IFNg and TRAIL In order to demonstrate the resistance of these cells to TRAIL at the mitochondrial level, we examined the premitochondrial enzyme caspase 8 activity in the TRAIL-mediated apoptotic signal pathway. Our data showed that caspase 8 was cleaved by TRAIL or by the combination of IFNg and TRAIL, appearing as an 18 kDa band corresponding to its active form (Figure 4a, lanes 3, 4). In contrast, caspase 8 was not cleaved by IFNg alone (Figure 4a, lane 2). To further address whether apoptosis is induced by IFNg and TRAIL through the mitochondrial pathway, we examined the release of cytochrome c from the mitochondria to the cytosol. We found that the combination of IFNg and TRAIL caused the concentration of cytochrome c to decrease in the mitochondria and to increase in the cytosol (Figure 4b, lane 4). This change did not appear with TRAIL treatment alone, despite activation of caspase 8 (Figure 4b, lane 3). Actin was used as a loading control in all experiments. Role of Bak in IFNg sensitizing TRAIL-induced apoptosis In order to clarify whether overexpression of Bak can sensitize ARO cells to TRAIL-induced apoptosis, ARO cells cotransfected with Bak and pcf1-luciferase plasmids were treated with TRAIL alone or combination with IFNg. The efficiency of transfection was controlled by luciferase activity. As expected, the cells transfected with the Bak plasmid became sensitized to TRAILinduced apoptosis, while the mock-transfected cells were unaffected (Figure 5a). We next investigated whether reduction of Bak protein modified IFNg sensitization to

TRAIL-induced apoptosis. Several stable transformants of antisense Bak expression vector were established. Western blot analysis of Bak and Bax proteins in the wild type of ARO cells as well as transformants is shown in Figure 5b. Even with IFNg pretreatment, ARO/ASBak remained resistant to TRAIL treatment (Figure 5c). ARO/AS-Bak cells had a lower percentage of dead cells compared to ARO/mock after IFNg and TRAIL treatment (Po0.001). Three independent, stable clones of As-Bak transfected ARO cells responded similarly, which excludes effects due to clonal variation. These data together suggested that Bak expression level is involved in the apoptotic pathway sensitized by IFNg. Overexpression of Bcl-2 or Bcl-XL in ARO cells does not block IFNg sensitizing effect To elucidate whether overexpression of Bcl-2, a common antiapoptotic molecule, in ARO cells blocks IFNgsensitized TRAIL-induced apoptosis, we treated ARO/ Bcl-2 and ARO/mock cells with TRAIL and IFNg. Following treatment, ARO/Bcl-2 cells had a similar percentage of living cells compared to ARO/mock cells (Figure 5d), indicating that the increase in Bcl-2 level is unable to block IFNg-sensitizing cell to TRAIL-induced apoptosis. We also found that ARO cells with overexpression of Bcl-XL did not block IFNg from sensitizing the cell to TRAIL-induced apoptosis (data not shown). In vivo inhibition of thyroid tumor growth by IFNg and TRAIL These studies were performed using the ARO cell lines. SCID mice bearing subcutaneous ARO cell xenografts were randomized into four groups: PBS, IFNg, TRAIL and IFNg plus TRAIL. As shown in an in vitro study (Figure 1a), neither IFNg nor TRAIL alone exhibited an antitumor effect against ARO cells xenografts (Figure 6). The combination of IFNg and TRAIL did demonstrate a significant inhibition of tumor growth (Figure 6). The body weight of the mice remained stable and there were no significant differences among the treatment groups. This in part reflected no major toxicity of the combination treatment.

Discussion

Figure 4 Analysis the mitochondrial pathway in ARO cell apoptosis induced by IFNg and TRAIL. (a) Determination of the active form of caspase 8. ARO cell were incubated with IFNg and/or TRAIL, and the active form of caspase 8 was determined (10 mg in each lane: lane 1, untreated; lane 2, IFNg; lane 3, TRAIL; lane 4, IFNg þ TRAIL). (b) Detection of cytochrome c in cellular fractions. After treatment, protein from the mitochondrial and cytosolic fractions (20 mg in each lane: lane 1, untreated; lane 2, IFNg; lane 3, TRAIL; lane 4, IFNg þ TRAIL) was used to detect cytochrome c release from mitochrodia. Actin was used as a loading control

IFNg has been used as a therapeutic agent in a number of autoimmune diseases, cancers and illnesses associated with viral infection Yano et al., 1996; Ossina et al., 1997; Beatty and Paterson, 2001; Shin et al., 2001). Previous studies have demonstrated that IFNg modulates cell death by sensitizing cells to proapoptotic but not pronecrotic stimuli (Ossina et al., 1997). Apoptosis induced by IFNg is documented in a number of human cancer cell lines including colon adenocarcinoma and hepatocellular carcinoma cells (Yano et al., 1996; Ossina et al., 1997; Shin et al., 2001). However, the mechanism responsible for IFNg-related apoptosis is not clear. Oncogene


932

Figure 6 Inhibition of thyroid tumor growth by IFNg and TRAIL. A cell suspension of 1 106 was implanted s.c. in the right flank of scid mice. When the tumors had reached a volume of 30–60 mm3, the animals were randomly allocated into three control groups or a treatment group. The controls received either 100 mg of TRAIL, or 100 U of IFNg, or 100 ml PBS, while the treatment group received a combination of 100 mg of TRAIL and 100 U of IFNg. All were administered by intratumor injection at days 0, 2, 9 and 14. The tumor growth was observed twice weekly by serial caliper measurements

Figure 5 Apoptotic responses in cells expressing Bak, antisense Bak or Bcl2 plasmid. (a) Effect of transient expression Bak in ARO cells. After 36 h of the transfection, the ARO/Bak and ARO/Mock cells were treated with TRAIL or combination with IFNg. (b) Western blot showed Bak, Bax and actin protein concentrations on wild-type ARO cell (lane 1), ARO/Mock (lane 2), and ARO/ASBak (lane 3). (c) Effect of stable expression ASBak in ARO cells. ARO/ASBak and ARO/ Mock cells were treated with TRAIL or combination with IFNg. (d) Effect of stable expression Bcl2 in ARO cells. ARO/Bcl2 and ARO/ Mock cells were treated with TRAIL or combination with IFNg. Cell death was determined in (a), (c) and (d) by flow cytometry

Oncogene

Although one study suggested that thyroid carcinoma cells are sensitive to TRAIL-induced cell death in serumfree media (Poulaki et al., 2002), our prior work showed that normal thyroid cells and thyroid carcinoma cells are resistant to TRAIL-induced apoptosis in the presence of serum. Our present experiments demonstrate that IFNg significantly increases the sensitivity of human thyroid cancer cells to apoptosis induced by TRAIL even in the presence of serum, while normal thyroid cells are resistant to this action of IFNg under the same conditions. This differential action of IFNg on apoptosis in thyroid cancer cells and thyroid normal cells may be clinically important, as it enables IFNg to specifically target the cancer cells for drug-induced apoptosis without affecting the normal cells. The present study reiterates the evidence that IFNg does not affect normal cells in the same manner as cancer cells in the human thyroid (de Metz et al., 2000). Soluble human recombinant DR5 could completely block IFNg sensitizing ARO cells to TRAIL-induced apoptosis, indicating that the death signal was specifically provided by TRAIL and not a contaminant or other mediator. IFNg can exert its effect on several molecules involved in the common cell death pathway in a shared or a celltype-specific manner (Fulda and Debatin, 2002). Many mitochondrial molecules are part of the common apoptotic pathway and play a critical role in causing apoptosis that is induced by either receptor or nonreceptor triggers (Kroemer, 2001). Our results demonstrate the resistance of thyroid cancer cells to TRAIL at the mitochondrial level since TRAIL treatment alone causes the activation of the premitochondrial enzyme caspase 8, but does not trigger apoptosis. Bcl-2 family members are apoptosis-related molecules that associate with other family member molecules in the mitochondria to regulate apoptosis. These proteins interact with each other to modulate the apoptosis threshold for the


933

cell (Kluck et al., 1997; Newton and Strasser, 1998). Our data showed that IFNg significantly increased Bak concentrations but had no effect on other Bcl-2 family members, such as Bcl-2 and Bcl-XL. The change in the ratio of Bcl-2 family members may influence how a cell responds to TRAIL. The involvement of Bak in facilitating apoptosis sensitized by IFNg was confirmed by the antisense Bak-specific blocking experiment. Application of antisense Bak effectively reversed the sensitizing effect of IFNg such that thyroid cancer cells remained unresponsive to apoptotic inducers. In addition, overexpression of Bak can sensitize ARO cells to TRAIL-induced apoptosis. These data together suggested that Bak is crucial to IFNg facilitation of apoptosis by TRAIL. It has been noted that Bak plays a key role in the formation of mitochondrial pores, which allow the release of cytochrome c and Smac, a mitochondrial protein (Shimizu et al., 1999; Sundararajan et al., 2001). Smac promotes caspase 9 activation by binding to inhibitor of apoptosis proteins, IAPs, and removing their inhibitory activity. This process leads to activate caspases and the cell death pathway (Shimizu et al., 1999; Sundararajan et al., 2001). Our data demonstrate that the combination of IFNg and TRAIL causes the release of cytochrome c from the mitochondria into the cytosol. This is correlated with an increase in the level of Bak caused by IFNg. Upregulation of Bak by IFNg could act through facilitation of cytochrome c and Smac release, and subsequent activation of caspases to sensitize the cell to TRAILinduced apoptosis. We also checked whether overexpression of Bcl-2 or Bcl-XL could block the cell death. ARO cells with overexpression of Bcl-2 or Bcl-XL did not alter IFNg sensitizing TRAIL-induced cell death, suggesting that the susceptibility of human thyroid cancer cells to apoptosis by IFNg via Bak production could not be altered by the overexpression of Bcl-2 or Bcl-XL. Although Bcl-2 or Bcl-XL is able to form heterodimers with the proapoptotic Bcl-2 members, Bax and Bak, it does not complex efficiently with Bak (Shimizu et al., 1999). Furthermore, it has been reported that Bak can accelerate apoptosis independently of its heterodimerization with either Bcl-2 or Bcl-XL (Simonian et al., 1997). This may explain why Bcl-2 or Bcl-XL does not block IFNg sensitizing TRAIL-mediated apoptosis. IFNg shares many of its biological effects with the type I IFN, IFNa and IFNb, and both IFNa and IFNb may stimulate apoptosis in certain types of cells, IFNa and IFNb were less effective in sensitizing thyroid cancer cells to the apoptotic inducers in our present experiment. This is contradictory to results reported with breast cancer cells (Kumar-Sinha et al., 2002), and it may indicate that the effect of IFNg in thyroid cancer cells is unique. A signaling crosstalk between the two types of IFNs in caveolar membrane domains has been proposed as a molecular basis for their overlapping functions and synergism (Sato et al., 2001). In view of our results, it seems unlikely that such a crosstalk functions in human thyroid cancer cells to promote apoptosis. In fact, type I IFNs has been reported to

inhibit apoptosis or stimulate growth and survival in some situations (Egle et al., 1996; Jelinek et al., 1997). In conclusion, the present study demonstrates that IFNg increases the susceptibility of human thyroid cancer cells to apoptosis mediated by TRAIL and that normal human thyroid cells are not affected by this treatment. This differential effect of IFNg on the thyroid cancer cells and normal thyroid cells may have therapeutical significance. In vivo experiments support the clinical value of this manipulation. We conclude based on our data that one of major mechanism of IFNg sensitization to TRAIL-induced apoptosis in thyroid carcinoma cells is through upregulating Bak expression.

Materials and methods Cell culture and treatments The ARO cell line, the FRO human cell line, and the NPA cell line were generously provided by Dr JA Fagin (University of Cincinnati, Cincinnati, OH, USA). These cells were grown in RPMI 1640 supplemented with 10% FBS, 1 nonessential amino acids and 1.0 nM sodium pyruvate. Primary normal and cancer thyroid cells were cultured in Cellgro media (Mediatech, Herndon, VA, USA) with 20% NuSerum IV (Collaborative Biomedical Products, Bredford, MA, USA). The purity of thyroid cell population was verified by staining with anticytokeratin 18 antibody (a marker for epithelial cells), quantitated by flow cytometry. Only cultures that were 490% cytokeratin positive were used for experiments. The cells were pretreated with IFNa, b and g (Roche Molecular Biochemicals, Mannheim, Germany) for various periods of time, and then were incubated with recombinant TRAIL (a gift from Dr A Chinnaiyan) for 16 h. In some experiments, soluble, recombinant human DR5 (a gift from Dr Y Chen) was added to cells 30 min before TRAIL administration. Quantization of cell death Cells treated with goat polyclonal anti-human DR5 antibody (R&D Systems, Minneapolis, MN, USA) or recombinant TRAIL were analysed for cell death by staining the cells with fluorescein diacetate (FDA) and propidium iodide (PI), with subsequent evaluation by flow cytometry as described by Killinger et al. (1992). In total, 10 000 cells were acquired from each sample for analysis. Assay of gene expression by cDNA microarray analysis The total RNA was isolated from cells using Trizol Reagent according to the manufacturer’s protocol (Molecular Research Center, Inc., Cincinnati, OH, USA). Isolated RNA was further purified by using RNeasy spin column (Qiagen Inc., Valencia, CA, USA), and then used to generate cRNA probes. Preparation of cRNA, hybridization and scanning of the Human Genome U95A Arrays was performed according to the manufacturer’s protocol (Affymetrix, Santa Clara, CA, USA). RNase protection assay RNA was isolated from cells using Trizol Reagent as mentioned above. The expression of Bcl-2 family genes was then evaluated by RiboQuant MultiProbe RNase protection assay system (Pharmingen, San Diego, CA, USA). [32P]labeled Oncogene


934 antisense RNA probe was prepared using the hAPO-2 (Bcl-2 family gene) template set according to the manufacturer’s protocol. Relative amounts of RNA loaded were corrected by comparison with the glyceraldehyde-phosphate-dehydrogenase (GAPDH) band intensity for each sample.

concentration by Western blot analysis and also were tested for apoptosis induced by TRAIL. We also established ARO cells with Bcl-2/pCDNA3 plasmid containing an N-terminal flag epitope tag (ARO/Bcl-2) or pCDNA3 plasmid (ARO/ mock) as described in Wang et al. (2001). These cells were used to test whether what can alter the IFNg sensitizing effect.

Western blot analysis After treatments, as described above, cells were lysed in Triton X-100 buffer as described previously (Wang et al., 1999) for the analysis of the Bcl-2 family of proteins or Chaps buffer (50 mM Pipes/HCl, pH 6.5; 2 mM EDTA, 0.1% Chaps, 5 mM DTT) for the caspase family of proteins. In some experiments, both adherent and floating cells were collected and used to prepare mitochondrial and cytosolic fractions as described by others (Kandasamy et al., 2003). Mouse anti-human Bak, Bid, Bcl2, Bclx and cytochrome c antibodies (Pharmingen, San Diego, CA, USA) were used according to the manufacturer’s protocol. Mouse monoclonal anti-caspase 8 was provided by Cell signaling Technology (Beverly, MA, USA). The cell lysates were separated by SDS–PAGE and then transferred to nitrocellulose membranes. After blocking nonspecific binding, the blot was incubated with the primary antibody and then an HRP-conjugated second antibody. To ensure equal loading and transfer, membranes were probed for actin using anti-actin (Ab-1) antibody (Oncogene Research Products, San Diego, CA, USA). Immune complexes were detected with an enhanced chemiluminescence detection method (Amersham, Arlington Heights, IL, USA). Transfection of thyroid cancer cells with Bak, antisence Bak or Bcl-2 expression vectors The complete coding sequence of Bak or antisense Bak was kindly provided in an expression vector (pRC/CMV) by Dr H Eguchi (Radiation Effects Research Foundation, Hiroshima, Japan). To perform the transient transfection, ARO cells were cotransfected with Bak plasmid and pcf-luciferase plasmid in 5 : 1 ratio using Fugene 6 (Boehringer Mannheim, Indianapolis, IN, USA) for 5 h. After 36 h of the transfection, the cells were treated with TRAIL or combination with IFNg. The luciferace activity and the percentage of cell death were analysed. To establish stable transformants, ARO cells were transfected with antisense Bak expression vector using Fugene 6, then selected in media containing 400 mg/ml G418 for 2 weeks. Isolated clones were trypsinized in plastic rings, and the cells were cultured in the presence of 200 mg/ml G418. Selected clones were analysed for alternations in Bak at protein

SCID mouse xenograft model In the experiment, 6-week-old female SCID mice from Charles River were used. A subcutaneous ARO xenograft tumor model was established. A cell suspension of 1 106 was implanted subcutaneously in the right flank. When the tumors were clearly palpable and had reached a volume of 30–60 mm3, the animals were randomly allocated into control or treatment groups and five animals were included in each group. The tumor growth was observed twice weekly by serial caliper measurements. Tumor volume was calculated using the formula (length width2)/2, where length (a) is the largest dimension and width (b) the smallest dimension perpendicular to the length (a b2)/2 (Lee et al., 1988). A combination of 100 mg of TRAIL and 100 U of IFNg was administered by intratumor injection at days 0, 2, 9 and 14 in amount of 0.1 ml. TRAIL, IFNg or PBS alone was also used as control. All of the animals were sacrificed 17 days after the last therapeutic injection. Tumors were removed for histology evaluation. Data analysis Flow cytometry data were analysed by WinMDI 2.8 (Joseph Trotter URL http://facs.scripps.edu/). Microarray data analysis was performed using GeneChip 4.0 software. Densitometric quantitation of autoradiograms was performed using Quantity One (Bio-Rad). Statistical analyses were performed using Student’s t-test, w2-test and Student–Newman–Keuls’ test. Acknowledgements We gratefully acknowledge Dr A Chinnaiyan for the endotoxin-free human recombinant TRAIL, Dr Y Chen for the soluble recombinant human DR5, Dr JA Fagin for the ARO, FRO NPA human thyroid carcinoma cell lines, Dr H Eguchi for the antisense Bak expression vector (pRC/CMV), Dr J Bretz and Dr N Beeson for comments on the manuscript. This work was supported by the National Institutes of Health Grants R01 A137141, P60DK20572 and DK58771.

References Ain KB. (1998). Thyroid, 9, 715–725. Ain KB, Egorin MJ and DeSimone PA. (2000). Thyroid, 10, 587–594. Beatty GL and Paterson Y. (2001). Immunol. Res., 24, 201–210. Bonavida B, Ng CP, Jazirehi A, Schiller G and Mizutani Y. (1999). Intl. J. Oncol., 15, 793–802. Braga-Basaria M and Ringel M. (2003). J. Clin. Endocrinol. Metab., 88, 1947–1960. Bretz JD, Arscott PL, Myc A and Baker Jr JR. (1999a). J. Biol. Chem., 274, 25433–25438. Bretz JD, Rymaszewski M, Arscott PL, Myc A, Ain KB, Thompson NW and Baker Jr JR. (1999b). J. Biol. Chem., 33, 23627–23632. Caturegli P, Hejazi M, Suzuki K, Dohan O, Carrasco N, Kohn LD and Rose R. (2000). Proc. Natl. Acad. Sci., 97, 1719–1724. Oncogene

de Metz J, Romijn JA, Endert E, Corssmit EP and Sauerwein HP. (2000). Thyroid, 10, 87–91. Egle A, Villunger A, Kos M, Bock G, Gruber J, Auer B and Greil R. (1996). Eur. J. Immunol., 26, 3119–3126. Fisher DE. (1994). Cell, 78, 539–542. Fulda S and Debatin KM. (2002). Oncogene, 21, 2295–2308. Jelinek DF, Aagaard-Tillery KM, Arendt BK, Arora T, Tschumper RC and Westendorf JJ. (1997). J. Clin. Invest., 99, 447–456. Kandasamy K, Srinivasula S, Alnemri E, Thompson C, Korsmeyer S, Bryant J and Srivastava R. (2003). Cancer Res., 63, 1712–1721. Kayagaki N, Yamaguchi N, Nakayama M, Kawasaki A, Akiba H, Okumura K and Yagita H. (1999). J. Immunol., 162, 2639–2647.


935 Killinger Jr WA, Dorofi DB, Tinsley Jr EA, Keagy BA and Johnson Jr G. (1992). Ann. Thorac. Surg., 53, 472–476. Kluck RP, Bossy-Wetzel E and Green D. (1997). Science, 275, 1132–1136. Kroemer G. (2001). Bull. Acad. Natl. Med., 185, 1135–1142, discussion 1143. Kumar-Sinha C, Varambally S, Sreekumar A and Chinnaiyan AM. (2002). J. Biol. Chem., 277, 575–585. Langaas V, Shahzidi S, Johnsen JI, Smedsrod B and Sveinbjornsson B. (2001). Anticancer Res., 21, 3733–3738. Lee YS, Bulliard DE and Zalutsky MR. (1988). Cancer Res., 48, 559–565. Mitsiades N, Poulaki V, Tseleni-Balafouta S, Koutras DA and Stamenkovic I. (2000). Cancer Res., 60, 4122–4129. Newton K and Strasser A. (1998). Curr. Opin. Genet. Dev., 8, 68–75. Ossina NK, Cannas A, Powers V, Fitzpatrick PA, Knight JD, Gilbert JR, Shekhtman EM, Tomei LD, Umansky SR and Kiefer MC. (1997). J. Biol. Chem., 272, 16351–16357. Pan G, Ni J, Wei Y, Genetz R and Dixit VA. (1997). Science, 277, 111–113. Poulaki V, Mitsiades CS, Kotoula K, Tseleni-Balafouta S, Ashkenazi A, Koutras DA and Mitsiades N. (2002). Am J. Pathol., 162, 643–654. Ruiz-Ruiz C and Lopez-Rivas A. (2002). Biochem. J., 365, 825–832.

Sato K, Hida S, Takayanagi H, Yokochi T, Kayagaki N, Takeda K, Yagita H, Okumura K, Tanaka N, Taniguchi T and Ogasawara K. (2001). Eur. J. Immunol., 31, 3138–3146. Shimizu H, Narita M and Tsujimoto Y. (1999). Nature, 399, 483–487. Shin EC, Shin WC, Choi Y, Kim H, Park JH and Kim SJ. (2001). Cancer Immunol. Immunother., 50, 23–30. Simonian PL, Grillot DA and Nunez G. (1997). Oncogene, 15, 1871–1875. Strasser A, O’Connor L and Dixit VM. (2000). Annu. Rev. Biochem., 69, 217–245. Sundararajan R, Cuconati A, Nelson D and White E. (2001). J. Biol. Chem., 276, 45120–45127. Voigt W, Bulankin A, Muller T, Schoeber C, Grothey A, Hoang-Vu C and Schmoll HJ. (2000). Clin. Cancer Res., 6, 2087–2093. Wang SH, Koenig RJ, Giordano TJ, Myc A, Thompson NW and Baker Jr JR. (1999). Endocrinology, 140, 1649–1656. Wang SH, Phelp E, Utsugi S and Baker Jr JR. (2001). Thyroid, 11, 725–731. Wang SH, Bretz JD, Phelps E, Mezosi E, Arscott PL, Utsugi S and Baker Jr JR. (2002). J. Immunol., 168, 2470–2474. Yano H, Fukuda K, Haramaki M, Momosaki S, Ogasawara S, Higaki K and Kojiro M. (1996). J. Hepatol., 25, 454–464. Yeung SCJ, Xu G, Pan J, Christgen M and Bamiagis A. (2000). Cancer Res., 60, 650–656.

Oncogene